웹 서버의 동시적 요구 재배치 스케줄링 = A Concurrent Request Scheduling Strategy in Web Servers

국문 초록 (Abstract)

우리는 웹 서버에서 수행되는 concurrent request들이 요구하는 문서의 정보를 이용하여 순서를 바꾸거나, 동시에 수행되는 순서를 미리 알아내어 메인 메모리 캐쉬에 있는 요청에 우선 순위를 줌으로써 서버의 성능을 향상시켰다. 본 논문의 스케줄링 알고리즘을 적용하지 않은 웹 서버와 적용시킨 웹 서버의 처리율, 사용자 응답 시간을 비교해 보니 스케줄링 알고리즘을 적용한 웹 서버는 request들을 재배치해야 하는 오버헤드에도 불구하고 20% 이상 처리율이 우수하였다.

다국어 초록 (Multilingual Abstract)

The process scheduling of the concurrent requests is not considered on most of the conventional web server when it is highly loaded. In this paper, the scheduling strategy which provides the service by ordering the process of the concurrent requests has been proposed. The number of requests processed per second has been improved about 20% more than the conventional Web server by scheduling the processing order of the concurrent requests.
The throughput, of the web server, hit ratio of the cache memory and the response time of the clients are measured through simulations. The proposed scheduling has reduced the response time of the clients by 0.1 milliseconds on the average. The improvement of hit ratio of the cache memory has not been observed, but the throughput of the web server has been improved more than 20%. Therefore the overall performance of the server has been improved by the scheduling strategy proposed in this paper.